DeepSeek V4-Flash Review: 9/9 at 1/20 the Cost of Opus

DeepSeek V4-Flash is the efficiency tier of DeepSeek's V4 family - and in our own executed coding test it did something its price tag says it shouldn't: it matched Claude Opus 4.8's perfect 9-of-9 score, at roughly one twentieth of the cost, and it even out-scored its larger sibling V4-Pro. This review covers what V4-Flash is, the tested results (with the honest caveats), the real pricing as of our July 29, 2026 capture - DeepSeek raised Flash's rate by about 56% between our test run and that capture, and we show what it does to the numbers - and exactly when to reach for it.

What DeepSeek V4-Flash is

DeepSeek V4-Flash is the efficiency tier of DeepSeek's V4 family, released April 24, 2026. It's a Mixture-of-Experts model — roughly 284B total parameters with ~13B activated per token — with a native 1M-token context and open weights. The headline is price: it's one of the cheapest frontier-tier models you can call, and as we found, that cheapness doesn't cost you correctness on routine coding.

What we tested

We ran nine coding tasks — chosen to separate models, not flatter them — across 13 models, executing each answer against hidden tests. The tasks: two-sum, valid-parentheses, merge-intervals, Roman-to-integer, longest-common-subsequence, a nested-dict flatten, top-k words, a token-bucket limiter, and a CSV-line parser (quoted fields with escaped quotes). Same set as our coding-cost benchmark.

Results: 9/9, for cents

V4-Flash solved all nine — the same perfect score as Claude Opus 4.8 — at a tiny fraction of the cost. Here's the full tested field, by measured cost per 1,000 tasks (priced at the list rates of 2026-07-17, the run date):

Those cells are the July 17 run at the list prices of that day. DeepSeek has since raised V4-Flash from $0.09/$0.18 to $0.14/$0.28 (captured 2026-07-29), which is a uniform 1.56× on both input and output, so the same token usage now reprices to about $0.20 per 1,000 tasks. Opus 4.8, V4-Pro, and Qwen3 Coder Next are essentially unmoved.

The takeaway survives the repricing, but it is no longer the same takeaway: V4-Flash matched the most expensive model's correctness at ~1/20th of its cost ($0.20 vs $4.05), and Qwen3 Coder Next is now about half its price rather than a hair under it. On these tasks, paying frontier rates still bought nothing extra — but Flash's claim to be the outright cheapest correct answer is gone. If price is the only thing you are optimising, Qwen3 Coder Next at $0.10 is the one to beat.

The V4-Flash beats V4-Pro surprise

The counterintuitive result: the small Flash out-scored the big Pro. V4-Pro (the 1.6T-parameter flagship) managed 8/9 — it tripped on the CSV-line parser — while V4-Flash went 9/9, at roughly a quarter of Pro's cost ($0.20 vs $0.74 per 1,000 tasks once Flash is repriced at its 2026-07-29 rate; it was 1/6th at the prices the run billed at, and only Flash's price moved). Bigger isn't automatically better on bounded coding tasks. (More on that decision in our V4-Pro vs V4-Flash guide.)

The catch: speed

V4-Flash is cheap and accurate — not fast. It generates reasoning before answering (it spent ~568 reasoning tokens per task), so it averaged ~14.5 seconds per task, versus Opus 4.8's ~6.1s and Qwen3 Coder Next's ~7.0s. So the honest positioning is: if cost-per-correct-answer is what you optimize, V4-Flash wins; if you need snappy interactive latency, a non-thinking coder like Qwen may fit better.

Pricing & access

On OpenRouter (deepseek/deepseek-v4-flash): $0.14 input / $0.28 output per 1M tokens, captured 2026-07-29. That is up about 56% from the $0.09 / $0.18 our July 17 test billed at, and it closes a gap this article previously reported: OpenRouter no longer undercuts DeepSeek's first-party cache-miss rate, it now matches it. Cache hits on DeepSeek's own API still fall to roughly $0.0028/1M — a ~98% discount on repeated context, and a bigger deal now that the miss rate has moved. The weights are open, so self-hosting is possible, though the 284B MoE needs real hardware. Note: DeepSeek's legacy deepseek-chat / deepseek-reasoner aliases were deprecated on 2026-07-24 in favor of the V4 ids.

What that does to the numbers above: the same measured run reprices from $0.13 to about $0.20 per 1,000 tasks. Flash is still roughly 1/20th of Claude Opus 4.8 and about a quarter of V4-Pro — but it is now roughly twice the price of Qwen3 Coder Next, which scored the same 9/9. Prices at this tier move on a scale of weeks; re-check before you build a routing budget on any of these figures.

from openai import OpenAI
client = OpenAI(base_url="https://openrouter.ai/api/v1", api_key="YOUR_KEY")
resp = client.chat.completions.create(
    model="deepseek/deepseek-v4-flash",
    messages=[{"role": "user", "content": "Refactor this function..."}],
)

Who should use it

FAQ